Social Conventions

Kelly on dating
4 min readDec 31, 2018

The core of community engineering is designing and enforcing social conventions. Part of the design criteria is that those conventions should be similar to what people are used to already, at least so people can better guess, understand and follow them. But that’s the least important criterion and simultaneously the easiest and most problematic: If two people come in with differing expectations they may conflict horribly and one or both of them will need to be reeducated about what the locally expected behaviors are.

For most communities the goals of community engineering are cooperative: Keep everything as open and interconnected as possible while simultaneously limiting the negative impact of trolls. Immediate measures of engagement are reasonable proxies for how happy and productive the users are, with the notable counterexamples of ‘If it bleeds, it leads’ and the occasional video game addict.

Straight dating apps share having to deal with trolls, plus a whole bunch of catfish, Instagram models looking for followers, desperate assholes, and sex workers looking for clients. The sex workers part is particularly unfortunate, because FOSTA shut down their appropriate forums. Dating app vendors would be well advised to lobby for legislation re-legalizing them, to make the experience better for everyone.

What makes straight dating apps different is that they’re more of a competition than a cooperation. Users scramble for the best matches they can get, and the success of one person tends to squeeze out the success of others. A reasonable analogy is pro athletes. They all train and work as hard as they can to perform as well as possible, but if they all trained and worked the same but lesser amount they’d have about the same rankings and all be a lot less injured. The dirty little secret of the sports world is that it would be just as good of a spectator sport if the record for the 100 meters was 11 seconds instead of 9.6. (What matters for watchability is having a few standout athletes with just the right balance between advantage above and vulnerability towards the rest of the field. Brushing up against the limits of human potential tends to remove standouts. Having stringent but leaky drug testing regimes which a few athletes manage to sneak around creates them.) What pro athletic leagues do to try to minimize training is have an off-season to let everybody’s bodies heal. In principle athletes can (and some do) get a modest advantage by training during the off-season but many can’t afford that at all or find that it simply isn’t worth it at their level.

Swipe-based dating apps offer a similar reprieve to the off-season: Instead of the men composing custom tweets for all the women (or spamming them all with the same looks-like-it-was-custom message) they simply swipe. In the end the women wind up having about the same preferences they would otherwise, but with everyone putting in a lot less work and aggravation, and the men the women do get connected with are much happier and less frustrated albeit less good at spam, which is a contraindicator anyway.

Some people and apps moralize about how too many men swipe right on everybody. This is a symptom of bad community engineering. The vast majority of women shown to the men will never be shown to them, so spending any time at all thinking about whether to match them is a waste of time. Supposedly people like swiping and it increases engagement, but there’s also a time-honored tradition of not-for-profits giving people volunteer ‘jobs’ which are nothing but busy work just because people who do that are much more likely to donate. While this tactic may work, I’m not a fan. It would be far better to simply cut out the useless activity: Have women do all the initial swiping, and only show men the women who have already swiped right on them. Then the men would have a plenty small enough sets to swipe on that they’d naturally carefully evaluate all of them and swipe honestly. This would also be a big privacy boost to the women, because their profiles would only be visible to the relatively small set of men they swiped right on instead of being essentially public information.

The other big problem with swiping is that it’s gameable. Do you get a benefit from swiping left on most people and only right on the especially attractive ones? Or is it advantageous to swipe right on almost everybody? Indications are you can get an advantage sometimes, from some apps, using some strategies, but the vendors are predictably close-mouthed about it because they don’t want everybody gaming the system. Inevitably there will be a fair amount of gameability, because a lot of users want some amount of exclusivity and the system wants everybody to get some matches, which leaves the system open to getting gamed by users excluding all but their stretch goal candidates. But countermeasures can be put in place, and there should be a principle that if the app appears to be showing profiles matching certain criteria in random order it’s actually showing them in random order. There will be more discussion of gameability in later posts.



Kelly on dating

Writing thoughts on dating. Because it beats going to therapy.